David Williamson Shaffer and his colleagues are developing an influential approach to education and assessment that relies on the notion of “Epistemic Network Analysis.” They posit that a “profession or other socially valued practice” (e.g., engineering) has an “epistemic frame” that is composed of many facts, skills, values, identities, and other concepts that advanced practitioners link together in various ways. Thus you can diagram a professional’s epistemic frame as a network and measure it using tools that have been developed for measuring social networks. What nodes are most central? How dense is the whole network? How many clusters does it have?

One way to collect the data necessary for this kind of analysis is to ask a practitioner to write or talk about her work. Many of her sentences will invoke concepts and link them together. (“I did A because I knew that B.” “I recommend C because I believe in ethical norm D.”) By coding the text, one can produce a dataset that can be displayed and analyzed in network terms. As Shaffer and colleagues note, the graph is not the actual epistemic network; it is a representation of how the engineer’s mentality works under specific practical circumstances (Shaffer et al, 2009, p. 14).

If a profession is worthy, then learning its epistemic frame is desirable. As students experience a course, a project, or an internship, their epistemic frames can be diagrammed and quantified at regular intervals. The learners’ networks should grow more similar to those of advanced professionals. Measures of network structure can be used for “formative assessment” (giving feedback on what the student should study) and “summative assessment” (awarding a grade or credential).

I have posited that moral thinking is also an epistemic frame (to use Shaffer’s terminology). We hold many morally relevant ideas that we connect by various kinds of links, not just logical inferences but also causal theories, generalizations, analogies, etc. We can graph our own moral mentality as a network of ideas and connections. Moral learning means building a moral network map that resembles that of a good moral thinker. (I leave aside for now the question of whether good moral reasoning is related to good moral behavior.)

One can easily see that the moral network map of an average adult is more complex than that of a 2-year-old. It is uncontroversial that a toddler needs to learn to reason more maturely, in which case his network map will look more like yours and mine. But that leaves a lot of room for debate about what an ideal map looks like. Defining good moral thought is a normative, not an empirical, question.

To some extent, that is also true of engineering. It is not self-evident what makes a “good” engineer. However, as long as we assume that the profession is working reasonably well and fulfilling its social purposes adequately, then a “good” engineer is presumably a respected and successful one. We can identify such people empirically: they have high grades, awards, and responsible positions. Then we can diagram their epistemic frames and compare novices to exemplary professionals to assess their learning.

The situation is much harder with morality. We debate what specific moral concepts and relationships should be found on a person’s epistemic frame. For instance, should everyone’s graph show the existence of God, linked to a set of commandments? We also debate what formal properties any moral network should display. Should it be highly centralized around one fundamental truth? Classical utilitarians and some religious fundamentalists would say so. Or should it be very flat and complex, as certain liberals have held?

Here I would introduce a controversial–but not original–premise that makes the identification of good moral networks somewhat more empirical. No human being can have a fully adequate moral theory in place before she faces the various situations of life. The moral world is far too complex for that. It involves countless differently situated people interacting in countless situations in relation to institutions (like education, romance, politics, and punishment, to name a few) that have evolved to have manifold purposes and meanings. So to think well morally is not to apply a theory to each new case, but rather to learn constantly. Learning results from interactions with other people (whether face-to-face or vicariously). By “interaction,” I do not mean only communication, or the exchange of ideas. Groups of people can agree on thoroughly foolish ideas unless they try to put them into practice. So “interaction” means a combination of exchanging ideas, trying to work together, and reflecting on the results–what Dewey often called “conjoint activity.”

Who is good at that? This is not strictly an empirical question, because we might disagree about how to assess various styles of interaction. Should we admire the persuasive ideologue? The follower of fads? But although value-judgments are inescapable, I think it is partly an empirical question who participates constructively in conjoint activity. Good participants do not impose preexisting ideas and do not merely adopt the majority’s view, but shape the group’s beliefs while adjusting their own.

As I have written before, my own unsystematic observation suggests that people who are better at moral interaction have epistemic networks with these features:

1. Lots of nodes and links, because each idea is an entry point for dialogue, and each reflects some prior learning.
2. A degree of centrality, because some moral ideas are genuinely more important than others; and also because one should develop a set of prized values that constitute your character. Yet:
3. No outright dependence on a small set of nodes to hold the whole network together, because then disagreement about those nodes must end a conversation, and doubt about them will plunge you into nihilism. You may believe in fundamental principles, but you should be able to reason around them. The network should be robust in that sense.

We might try to identify the actual epistemic frames of people who are good at collaboration and deliberation and see if they manifest the three features I listed above. We could then map the networks of children and other moral learners to see if they are developing to resemble the exemplary cases. Again, this would not be a value-neutral research program, but it would have a strong empirical component.

We can, in fact, pursue three levels of analysis.

1. Each individual has an evolving and not-fully-conscious epistemic frame composed of many ideas and connections.
2. The individual belongs to a community of other people who all have networks of their own. Their networks overlap and influence each other because moral learning is social. (Even a recluse got his ideas from someone else). Within a community, individuals’ maps intersect in a second way as well. If one person has a moral commitment to a specific other person, that other will appear on her map.
3. Finally, the world is composed of many moral communities. But these are never fully separate and distinct. They are always complex, overlapping, and vaguely-bordered networks. Given two entities that we call “cultures,” no matter how remote, we will likely find common nodes and connections in their respective moral networks. I leave aside the possibility that all human beings share a set of ideas as our biological inheritance. That may be the case, but I do not rely on it. Rather, all communities interact (even the so-called “uncontacted peoples” who live deep in rain forests), and so the members of community A always share some nodes with members of community B nearby as a result of their “conjoint activity.”

At the individual, community, and global level, the process of moral reasoning is fundamentally the same. It is always a matter of developing a more satisfactory network of ideas and connections. This is not easy, conflict-free, or pretty. Individuals face deep internal conflicts among incompatible ideas, and people and communities often actually kill each other on account of such disagreements. Nevertheless, we can point to individuals and groups that are better at constructive engagement, and moral learning means becoming more like them.

Reference: David Williamson Shaffer, David Hatfield, Gina Navoa Svarovsky, Padraig Nash, Aran Nulty, Elizabeth Bagley, Ken Frank, Andre A. Rupp, and Robert Mislevy, “Epistemic Network Analysis: A Prototype for 21st Century Assessment of Learning,” International Journal of Learning and Media, vol. 1, no. 2 (2009), pp. 1-22.